Arxiv:1610.07839V3 [Gr-Qc]

Home , Schwarzschild radius

arXiv:1610.07839v3 [gr-qc] 26 Jun 2017 mtigpeHwigrdainbfr h vn oio is in horizon necessity logical event no the is before there radiation then pre-Hawking Bob), emitting distant the to ing the restore can Both not. do they finite order. or a causal — within time horizon Bob’s event of defined coll amount suitably finite-time a a of facilitate crossing — effects quantum either lapse: a time two evaporation subsequent these i Bob. finite by of contradiction: perceived consistency vs. causal time on apparent collapse focus an finite We is there 9–13]. since elements 5, pa 4, the for [2, ingredients dox provide cor evaporation of the non-restoration) during alleged tions the mat (or collapsing restoration the and by ter horizon event the of Crossing problem. describe to way a a geometry. of provide spacetime conclusion and changing a the horizon to the lead of area holes decrease horizon black steady event stationary between for mass relationships and the conserva and infinity, energy, Positivit at of flux matter. energy steady-state collapsing emitte resulting the the the of from field(s) distinct (iii) analysis; are the diation of stage le first at neglected, the is at geometry spacetime the backreaction on the matter typically (ii) created each background; classical 9], a gravitational as the [3–5, treated (i) is ways [9]: assumptions of basic number three radiati on a hole based in Black derived evaporates. be time Bob’s hole can of black amount the finite a until descr only elapses are 8], effects 5, ra- [4, late-time formula Hawking its Page’s If by and [7] [2–5]. exists holes [3–6] black diation of models quantum in role relativ general classical of time. results of [1]. standard amount finite the a are only co- These lasts the to it According (Alice) escape. observer spacetime can moving the signal of no collapse boundary which a — from of horizon region stage event final the The of crossing time. is of blac amount a infinite into an distribution takes matter a of collapse infinity), tial fqatmefcsla oafiietm olpe(accord- collapse finite-time a to lead effects quantum If col- the of stages final the for alternatives two are There information hole black the precipitates radiation Hawking key a plays also horizon event an of notion the However spa- at (Bob observer distant a of clock the to According itnefo hswudb oio htdpnsol nthe wi on flux only and depends density energy that firewall-like horizon encounters observer I would-be crossed. this not from sh is distance the radius for Schwarzschild motion the of but equation evaporation, modified re a the in du and results thin metric Vaidya hypothesis massive outgoing evaporating the of geometry: collapse exterior the the finit of a context in the matter in collapsing gravitationally of crossing zon eivsiaetepsiiiyta unu fet respon effects quantum that possibility the investigate We .INTRODUCTION I. 2 1 eateto hsc n srnm,Uiest fWaterlo of University Astronomy, and Physics of Department eateto hsc srnm,McureUiest,S University, Macquarie Astronomy, & Physics of Department 3 eiee nttt o hoeia hsc,Wtro,On Waterloo, Physics, Theoretical for Institute Perimeter aetn Baccetti, Valentina oeo vprto ngaiainlcollapse gravitational in evaporation of Role 1 oetB Mann, B. Robert hole k fthe of ra- d rela- apse field ibed of y tion ast ra- ity on n- s - iea enb itn bevr.W osdrthis consider We observers. distant by seen as time e oia osqecs[1 hudntasm hmeither. them assume as- not not should [11] do o consequences investigations radiation logical [3–6], Hawking structures the background exotic everything of sume from derivations “original” differently Since the behaves or else. reason time, some coordinate for finite matter a the th in cross particle also horizon non-classical. will test formed matter be a collapsing the still either behind would that just trails hole implies speed-up black collapse large The a hor even the of containing region zon spacetime a Nonetheless, formed. eesr acltoa eal.W use We t all details. provide gener calculational Appendices IV. necessary applicability, Section shell their in implications thin of and a ity discussion examples. of the detailed aspects with several classical conclude presents III the Section review assump as collapse. specific discuss well their we as II models, Section tions our In to realized. leading is considerations — horizon no p — second sibility the where scenarios consistent several present and of 22–24]. loss [20, arises alleged throug never the 21], disappearance case 10, its 5, latter of [4, sense the horizon a In in least not. at do time information, they finite crossin or in horizon case) Bob the quantum for the for in allow defined radiation properly (however hole resp black effects quantum for either sible alternative: same the presents idea. a this [16] support shells [17–20] collapsing results of analytic studies Numerical made [13–15]. been matter fore have collapsing radius, gravitational the the when onset near begins the concentrates with that overlaps radiation and ac Hawking complete collapse of never the is that Bob Arguments to cording cl radius. sufficiently Schwarzschild is its matter collapsing to the when start must tion ftemti n set and metric the of l ei ihqatmfil hoyo uvdbackground [25] curved gravity a stochastic on semiclassical through theory advance field and quantum with mod- begin Relevant gravity. els and mechanics quantum between play sedteseli lasa eti sub-Planckian certain a at always is shell the nstead hantrlcutoff. natural a th il o lc oerdaind o lo o hori- for allow not do radiation hole black for sible addShazcidmti.W eciehwthis how describe We metric. Schwarzschild tarded fqatmefcsd o aiiaesc olpe evapora collapse, such facilitate not do effects quantum If eepoetecneune fpeHwigradiation pre-Hawking of consequences the explore We evaporation and collapse between overlap the Accepting hr saueu irrh fmdl htdsrb h inter the describe that models of hierarchy useful a is There tsel sn w aiiso erc odescribe to metrics of families two using shells st l.I ahcs h olpei ceeae u to due accelerated is collapse the case each In ell. ,3 2, asadeaoainrt,wieacomoving a while rate, evaporation and mass n ailR Terno R. Daniel and I EEA CONSIDERATIONS GENERAL II. ,Wtro,Otro Canada Ontario, Waterloo, o, de S 19 Australia 2109, NSW ydney ai,Canada tario, c = ~ = 1 G = k B 1 = ( − . ++) + signature fully its f and ose on- We be- os- a h al- nd he at i- g - - - - 2 different effective field theories of matter-gravity systems [26] where g = det gµν and W is the effective action of quantum to a full theory of quantum gravity in whatever form it takes fields [3], but only partial results are known for dynamical [2]. Black holes are discussed at each level of this hierarchy. spacetimes. Appearances of strongly non-classical regions, both deep Below we illustrate the consequences of the Assumptions within appropriately defined quantum black holes and outside 1-4 in two examples. We discuss a generic spherically- the Schwarzschild radius [9, 27, 28] are expected. Neverthe- symmetric setting in [30]. less, the classical event horizon and quantum states that are associated with it play an important role in the quantum black hole models. Moreover, whatever reservations can be raised A. Themodel against the semiclassical picture of black hole creation, radiation and evaporation, they are still the only ones that are fully solvable. Hence investigations of the event horizon crossing We consider collapse of a massive spherically-symmetric thin dust shell Σ [31] in D+1 dimensional spacetime, D 3. within the semi-classical framework are important to establish ≥ its consistency and delineate its limitations. Appendix A summarizes the relevant conventions and defini- tions. Our treatment of this model is based on the following assumptions: The spacetime inside the shell is flat [31]. In the absence of Hawking radiation the metric

1. The classical spacetime structure is still meaningful and 2 2 −1 2 2 ds+ = f(r+)dt+ + f(r+) dr+ + r+dΩD−1 (3) is described by a metric gµν . − 2 2 = f(r+)du+ 2du+dr+ + r+dΩD−1, (4) 2. Classical concepts, such as trajectory, event horizon or − − singularity can be used. where f(r) =1 C/rD−2 describes the exterior geome- − try in terms of standard coordinates (t+, r+) and outgoing Assumption 1 leaves out the spacetime fluctuations. Within Eddington-Finkelstein coordinates (u+, r+). Here and in the this framework there are no restrictions on the applicability following dt+ = du+ + dr+/f(r+), and dΩD−1 is the spher- of classical concepts even, e.g., within the sub-Planckian dis- ical volume element. Their counterparts in the Minkowski tance from the Schwarzschild radius. While this is a stronger spacetime inside the shell are (t−, r−) and (u−, r−), where 1/(D−2) assumption than just validity of the classical description out- u− = t− r−. The Schwarzschild radius rg := C side the stretched horizon [4, 5, 27], it is consistent with the [32] is the− solution of f(r)=0. In three spatial dimensions goal of testing accessibility of an infinitely sharp classical C =2M = rg. event horizon. Moreover, while this assumption is not usu- The shell’s trajectory is parameterized by its proper time τ ally emphasized, it underlines many discussions of the Hawk- (time of a comoving observer Alice) as T±(τ), R±(τ) or ing radiation and related problems. Specifically, the standard U±(τ), R±(τ) using, respectively, (t, r) or (u, r) coordi- Penrose diagram of the creation of a black hole from the in- nates outside and inside the shell. falling matter with subsequent evaporation makes sense only The first junction condition [31, 33], which is the statement if both Assumptions 1 and 2 are accepted [29]. that the induced metric h is the same on the both sides of We furthermore assume ab the shell Σ, ds2 = h dyadyb = dτ 2 + R2dΩ − , leads to Σ ab − D 1 3. The collapse leads to a pre-Hawking radiation. the identification R+ R− =: R(τ). Trajectories of shell’s particles are timelike,≡ so 4. The metric is modified by quantum effects. The result- 2 ing curvature satisfies the semiclassical equation T˙+ = F + R˙ /F, (5) p R 1 Rg =8π Tˆ , (1) where the dot denotes a τ derivative and µν − 2 µν h µν i where R is the Ricci tensor correspondingto the met- R˙ + F + R˙ 2 µν ˙ (6) ˆ U+ = − , ric gµν and Tµν is the expectation value of the stress- pF energy tensor.h i where F = 1 C/RD−2. The surface stress-energy tensor Assumption 3 is the direct consequence of our discussion in for a thin dust shell− is Section I. In principle we make no assumptions about the spe- ab a b a b cific form of the pre-Hawking radiation. It should be obtained S = σv v = σδτ δτ , (7) as a part of a self-consistent analysis. The expectation value of the stress-energy tensor should include all matter fields as where σ is the surface density and va are the components of well as gravitons, thus relaxing assumptions (ii) and (iii) from the proper velocity in the surface coordinates ya. Discontinu- Section I. Formally it is given by ity of the extrinsic curvature Kab is described by the second junction condition [31, 33] 2 δW Tˆµν = , (2) h i √ g δgµν Sab = [Kab] [K]hab /8π, (8) − − − 3

a where K := K a, and [K] := K Σ+ K Σ− is the discontinu- we obtain ity of the extrinsic curvature K across| − the| two sides Σ± of the ˙ 2 ˙ 2 surface. The equation of motion for the shell can be written as ¨ 2R 1+ R dC −1 R = D−4 + (x ), 2 D−2 ˙ ˙ 2 2 dU O ′ (D 2) C p(R + 1+ R ) x 2R¨ + F R¨ − (12) (R) := p D 2 F + R˙ 2 − 1+ R˙ 2 where close to the Schwarzschild radius ˙ 2 ˙ 2 (D 2)x (D 2)x p F +pR 1+ R F − − . (13) +(D 2) − =0 (9) ≈ C1/(D−2) ≡ r − p R p g ′ ¨ where A˙ = dA/dτ, A = ∂A/∂r Σ. This equation is sim- The coefficient of dC/dU in (12) is positive, so R< 0. Hence ple enough to have an analytic solution| τ(R), leading to the sufficiently close to the Schwarzschild radius the collapse is finite crossing time τ(r ). Appendix B presents details of the always accelerated. At later stages (when R˙ & 2) it can be g | | derivation of the equations of motion. approximated as Unlike previous investigations of thin shell collapse [34– ˙ 4 36], we are not dealing here with its influence on a quantum ¨ 4R dC R D−4 < 0. (14) field, but rather focus on the effects of the resulting radiation ≈ (D 2)2C D−2 x2 dU on the shell dynamics. A metric describing the geometry out- − side the evaporating shell is self-consistently determined from We now evaluate the shell’s rate of approach to rg. In its vicinity Eq. (6) reduces to U˙ 2R/F˙ . Using this expres- the Einstein equations (Assumption 4). Using a null shell ≈ − [19, 28, 35] simplifies the analysis. Here we investigate a sion and Eq. (13) in the chain rule evaluation of r˙g we find massive shell, with the advantage of being able to consider that its evolution in the proper reference frame of Alice in addition 2rg drg dC to the asymptotic frame of Bob. The proper reference frame is x˙ R˙ 1 . (15) ≈ − (D 2)x dC dU crucial for establishing the physical meaning of the horizon. − We illustrate the general approach that was outlined in this Hence a natural time-dependent scale for this problem is section by two particular scenarios. The first scenario models 2 dC the geometry outside the shell using the outgoing Vaidya met- ǫ∗ := , (16) (D 2)2C(D−4)/(D−2) dU ric [19, 37–39], which is a popular model of exterior geometry − of evaporatingblack holes. Apart from two obviousconstrains and we do not specify the mass function C(u). Next we introduce the retarded Schwarzschild metric, which allows to describe x˙ R˙ 1 ǫ∗(τ)/x(τ) . (17) evaporation from the point of view of Bob. ≈ − By monitoring Then the gap decreases only as long as ǫ∗< x. If this is true for the entire duration of evaporation, we have R > rg until the evaporation is complete. Otherwise, once the distance be- x := R rg, (10) − tween the shell and the Schwarzschild radius is reduced to ǫ∗, where R(τ) is determined by the equations of motion based it cannot decrease any further. on the assumed outside metric, we analyse how evaporation From the point of view of Alice the collapse acceler- modifies the classical shell dynamics. ates, while for Bob the shell is still stuck within a slowly changing coordinate distance ǫ∗ from the slowly receding Schwarzschild radius. Furthermore using (6) it is straightfor- III. EXAMPLES ward to show that close to rg dC R˙ A. Vaidya metric C˙ 2C | | (18) ≈ dU x The outgoing Vaidya metric is given by (4) but with f and so Alice will see the evaporation rate C˙ vanish if and only f(u, r) = 1 C(u)/rD−2, where we only assume that→ if Bob does; if dC/du 0 then (11) reduces to Eq. (9), ǫ∗ − → → f(u, r) > 0 for r > rg(u) 0 and dC/du 0. Here and in 0, and an horizon forms. If dC/du < 0 then since the gap x the following we drop the subscript≥ “+” from≤ the variables. never vanishes, R˙ increases but remains finite for finite x. The equation of motion of the shell (9) becomes (see Ap- The only non-vanishing component of the stress-energy pendix B), tensor is 1 dC ˙ (19) R 1 8πTuu = 2 . (R) FU U˙ =0, (11) −r du D − ˙ 2 − 2F 2F F + R ! The energy density in the frame of an observer moving with a p four-velocity vµ is where Au := dA/du, and U˙ is given by Eq. (6). Solving µ ν Eq. (11) for R¨, then expanding in inverse powers of x and C ρ = Tµν v v , (20) 4 hence on the outer shell surface where const. We introduce C(t˜) via the implicit relationship (27), C(t˜) := CB (t r∗ + R∗) , which in particular implies U˙ 2R/F˙ 2RC/x˙ (21) − ≈− ≈− dC dCB(z) we have = . (28) dt˜ dz z=t˜ 1 dC R˙ 2 ρ . (22) It is straightforward to show (Appendix C) that for any choice ≈ 2π dU x2 of the function C this metric is inequivalent to the outgoing

The energy flux in this frame is given by Vaidya metric. In this model the shell follows the classical trajectory µ ν jn := Tµν v n = ρ, (23) (T (τ), R(τ)) until a certain coordinate distance ǫ from the Schwarzschild radius, when the evaporation abruptly starts. where nν is given by Eq. (B2).see that the observer Using While definitely not an instantaneous event, the ignition time Eq. (12) it is possible to see that Alice will experience an can be reasonably well-defined in, e.g., the adiabatic approx- enormously increased energy density and flux. Approximat- imation [42], and we shall set τ = t = 0 at this event for ing x ǫ∗ = 2C dC/dU the asymptotic form of the equa- convenience. Writing F := f˜(T, R), the equation of motion tion of∼ motion becomes| | (9) for the collapsing and evaporating shell now becomes (Ap- pendix D) R˙ 4 R¨ . (24) ≈ CC F R˙ U T (29) (R) 2 =0 Hence Alice will see the radiation flux growing to some maxi- D − F mal value contingent on the properties of the shell. This blue- where shift behaviour is a known feature of the standard black hole ′ C(D 2) dC ∂t˜ 1 models [40] and provides a possible realization of a firewall F = − , (30) RD−1 − dt˜ ∂r RD−2 acting via “internal conversion” [41]. It could be regarded as Σ a firewall with a natural upper cut-off. dC ∂t˜ 1 FT = . (31) − dt˜ ∂t RD−2 Σ

B. Retarded Schwarzschild metric The derivatives of the retarded time on the shell are ∂t˜ 1 ∂t˜ 1 We introduce an alternative modiﬁcation of the metric (3) = , = , (32) ∂t 1 R /F ∂r −F R that describes spherically-symmetric evaporation. This ex- Σ T Σ T − − ample demonstrates that horizon avoidance is not limited to Eq. (13 ) holds apart from the last stages of evaporation the Vaidya metric. We incorporate an mass function C(t) (as (where x(τ) C(τ)). Hence (5) becomes described by Bob), CB (t), into a local causality-respecting ∼ r metric via the retarded time, similarly to the derivation of the T˙ + g R˙ . (33) Lienard-Wiechert potential in the radiation problem [1]. Asa ≈ (D 2)x| | result we can illustrate the analysis by numerical simulation − of the process in Section IIIC. Noting that Outside the shell (the region ) we assume (t, r): r > R(t) dR dR R˙ the metric R˙ = T˙ , (34) dt ≈− dt F 2 2 −1 2 2 ds = f˜(t, r)dt + f˜(t, r) dr + r dΩ − . (25) + D 1 implies the equality R˜ = R R F , the evaporation − t|Σ t ≡ T ≈− This is the minimal modiﬁcation of Eq. (3) consistent with parameterized by the shell’s proper time is given by the assumption that an outgoing massless particle propagates dC dC ∂t˜ ∂t˜ in the Schwarzschild spacetime that is frozen at the moment = T˙ + R˙ (35) dτ dt˜ ∂t ∂r ˜ Σ Σ of its emission t, i.e. dC R˙ dC(T ) R˙ dr = (36) out ˜ D−2 ≈− dt˜ F − dT F =1 C(t )/rout . (26) T =T (τ) dt −

− since at short distances from the Schwarzschild radius Here f˜(t, r)=1 C(t˜)/rD 2 f(t,˜ r), and the retarded 1 ∂t/∂t˜ Σ and ∂t/∂r˜ Σ 1/(2F ). time t˜(t, r) is given− by the implicit≡ equation | ≈ 2 | ≈− Solving Eq. (29) for R¨ and expanding in inverse powers of r ′ x and C we obtain ˜ dr ˜ t t = ′ r∗ R∗(t), (27) − R(t˜) f(t,˜ r ) ≡ − ˙ 2 ˙ 2 Z ¨ R 1+ R dC −1 R = D−4 + (x ), where the tortoise coordinates are defined using the mass pa- 2D(D 2)2C Dp−2 R˙ + 1+ R˙ 2 x2 dT O rameter at t˜. Outgoing null geodesics are the lines of t˜ = − (37) p 5 for the dominant term of the shell’s acceleration as it approaches the Schwarzschild radius, and the expression further simplifies (Eq. (41)) for R˙ & 2. | | Approach to the shell x˙ = R˙ r˙g is still governed by Eq. (17). Using Eq. (36) we find that− now the distance scale is

1 dC ǫ∗(τ)= . (38) (D 2)2C(D−4)/(D−2) dT

−

As an illustration we consider the late-time evaporation law

dCB D 2 1 = − − , (39) dt − Dκ C2/(D 2) B FIG. 1. Example: D = 3, C0 = 10, R0 = 100. The exact solution (black dotted line) for the gap x(τ) = R(τ) − r (τ) ≡ R(τ) − where the constants are discussed in Appendix E. Starting g C(τ) of the system of Eqs. (29) and (35) is evaluated up to τ = from the initial value C0 the evaporation lasts a finite time 1.434846531778 (t ≈ 89.5697), where the numerical integration tE, breaks down. The steady state value ǫ∗ = 1/3κC is shown as the orange line. D−2 D C (t)= C (1 t/t ) D , t = κC D−2 . (40) B 0 − E E 0 Using (39) the asymptotic dynamics of the evaporating This problem is avoided if one rewrites the equations in shell is given by the system terms of Bob’s time t = T (τ), using ˙ ˙ ¨ ˙ 2 ¨ R˙ 4 R = RT T, R = RT T T + RT T. (43) R¨ , (41) ≈−D(D 2)κCx2 ˙ ¨ ˙ ˙ − Using Eq. (25) we obtain T , and T = (dT/dT )T (Ap- 1 R˙ pendix D), thus allowing us to replace the derivative of R over C˙ , (42) ≈ DκC1/(D−2) x the proper time in Eq. (29) by the derivatives over Bob’s time. The approximate equation for x takes a particularly simple where (41) is valid for large R˙ & 2. form We see that the effect of| evaporation| is negligible until dx dR drg (D 2)x drg about τ1 ǫ/ R˙ (0) , when the distance to the Schwarzschild = − (44) ∼ | | dT dT − dT ≈− rg − dT radius reaches x = (ǫ∗). Once it is reached we are O guaranteed that x˙ > 0 and so the shell does not cross the and can be also obtained directly from Eq. (17) by using the Schwarzschild radius despite rapidly increasing acceleration. value of F RT from Eq. (13). For D =3, ǫ∗ =1/3κC, While our choice of the switching-on parameters determines we find | | ≈ | | the value of τ1, the steady-state coordinate distance ǫ∗ de- dx x 1 pends only on the system properties. = + . (45) dt −C 3κC2 When Eq. (40) is substituted for C(t) this equation has a so- C. Numerical solutions lution in terms of the error function. The numerical solution to this equation that extends the exact solution to all times is We illustrate the situation by numerically solving (29) and depicted in Fig. 2. (35) using the evaporationmodel of Eq. (40). We take D =3, C0 = 10, and ǫ =1. We use κ = (5120π/8) 1/8,where the last factor of 1/8 converts from the usual three-dimensional× D. The core and the shell model expression for M = C/2. The evaporation time in this case is 5 tE =2.513 10 . It is convenientto assume that the collapse A more complicated scenario involves a massive started with× the shell at rest (in this example at R = 10C ), 0 0 spherically-symmetric core of radius Rc with the and simply vary the initial radial coordinate and the mass to 1/(D−2) Schwarzschild radius rc = C . Assume a shell cover all possible scenarios. g c collapses from rest at some R(0), such that initially The rapid increase of R¨ after τ τ1 leads to breakdown of the numerical integration| | of either∼ the exact or approximate c rg < Rc < rg < R (46) systems for sufficiently small x. Physically it means that after the shell comes to within a distance ǫ∗ from the Schwarzschild where C > Cc yields the total gravitating mass of the system. c radius, Alice sees the rest of the collapse and evaporation as For a large core (rg Rc) an observer that is situated any- nearly instantaneous: for the last point on the shell trajectory where between it and the≪ shell should not experience any de- on Fig. 1, time dilation factor is T˙ =3.05 107. viation from classical relativity. Hence the shell is the only × 6

the examples we presented, for almost the entire evaporation time the shell stays very close to its Schwarzschild radius, but never crosses it. As a result, there are no trapped surfaces, − no horizon and no singularity. The distance ǫ∗ C 1 grows as the shell evaporates and a comoving observer∝ encounters an increased radiation flux with a natural cutoff — a kind of tame firewall. The distance from rg is within the trans-Planckian regime, but no more so than in the usual derivation of Hawk- ing radiation. Both horizon avoidance and its trans-Planckian scale support the idea that the paradoxical aspects of the black hole information problem originate in combining a sharp classical geometry with quantum fields [29, 44]. It is quite reasonable that the mean field view of the collapse, radiation and backreaction breaks down well before the gap ǫ∗ is reached. Quantum fluctuations become important and should be taken into account at least at the level of FIG. 2. Approach of x(t) to ǫ∗ = 1/3κC (orange line) in the same the stochastic gravity [25]. However, following this line of ≈ setting. On this time scale ǫ∗ const. The plot is obtained as a thought still implies that a classical picture of matter crossing combination of the solution of the system of Eqs. (29) and (35) (for a sharply defined surface is untenable, while quantum states t< 89.5697, black dotted line) with the solution of the approximate that are associated with such a surface are not asymptotically Eq. (45) and with the adjusted initial conditions (t > 50, blue line). reachable. The inset (a) shows the approach of x(t) to ǫ∗(t). After the moment t∗ when x(t∗) = ǫ∗(t∗) the distance from the Schwarzschild radius Despite ignoring the fluctuations our simple model is self- begins to increase, but it trails behind ǫ∗(t). The inset (b) shows consistent. The examples we have considered can be general- the two scales of the problem: log C(t) (dashed dark red line) and ized to an arbitrary spherically-symmetric metric outside the log ǫ∗(t) (orange line) through the evaporation. shell [30]. Both avoidance of the horizon and a regularized firewall do not require any additional assumptions or exotic matter. source of Hawking radiation, and Eqs. (11) and (29) are eas- Currently popular approaches to preserve unitary dynamics ily generalized (Appendix F). Upon analysis we find that the that are based on the analysis of matter alone (such as firewalls horizon still is not going to be reached, and the outcome of [27, 48], final projections [49], or ER=EPR [50, 51]) require the evolution (the shell evaporates or crashes into the core) is both an horizon and a singularity and so are not applicable if determined by the two quantities: x(τ), and an event horizon does not form. Our results also indicate the standard Penrose diagram that is used to illustrate the black h(τ) := rg R . (47) − c hole creation and evaporation is inapplicable [29]. We first observe that in any dimension the shrinking Description of the entire spacetime in terms of a classical Schwarzschild radius will retreat inside the core, since at τr metric, even without the assumption that the Einstein classical equations are be violated in the deep Planck region [47], D−2 C(τr)= Rc > Cc, (48) results in disappearing of the event horizon, consitent with the expectations of [9]. While it is reasonable to assume indicating that the evaporation is not complete. Assuming that that avoidance of a suitably defined Schwarzschild radius is the shell is close to the Schwarzschild radius (and using the a generic feature of a quantum collapse, lack of trapped sur- retarded Schwarzschild metric) we approximate x(τ) as ǫ∗(τ) faces and apparent horizon is most likely a consequence of (Eq. (38)), and the condition for crash becomes the thin shell model. In general we expect appearance of the −1 quasi-locally defined trapping horizons that should enable to C1/(D−2)(τ )+ D(D 2)κC(τ ) = R . (49) c − c c contently discuss collapse, formation and evaporation of black holes [9]. The crash occurs if C(τc) > Cc. Otherwise the shell evapo- The smallness of ∗ seems to indicate that predictions of rates beforehand. For example, in D =3, the crash is possible ǫ if the core is large enough relative to its Schwarzschild radius, this model should be observationally indistinguishable from pure classical collapse [45]. However, it was recently shown

C(τc) Rc 1/3κRc > Cc. (50) that hypothetical non-black-hole very compact objects can ≈ − have a very different quasinormal-mode spectrum from that This set-up provides a complimentary scenario to Ref. [43]. of black holes, even in the limit of coinciding exterior metrics. This difference is not manifested in the ringdown signal from a binary coalescence, but may become detectable in precision IV. DISCUSSION observations of the late-time ringdown signal [46]. Investigation of the energy-momentum tensor and its com- Including the effects of pre-Hawking radiation dramatically parison with the results obtained from other considerations modifies the evolution of gravitationally collapsing matter. In [17–19, 34, 36, 38, 52, 53], as well as generalization to a non- 7 zero pressure are natural extension of our results and will be ture with the surface energy-momentum tensor. Here we cal- published elsewhere. culate the extrinsic curvature outside the shell. Expressions Our model strengthens the point of view (see [28, 35, 54, for the interior are analogous. Components of the extrinsic 55]) that fully quantized joint gravity-matter dynamics must curvature are given by have unitary time evolution, particularly for systems that have α β α a well-defined classical Hamiltonian, and so there cannot be Kττ := nαv v = nαa ,Kφkφk := nφk ;φk . − ;β − any overall information loss. Indeed, all current candidate the- (B1) ories of quantum gravity— strings, loops, and foams — are with k =1,...,D 1 , where vµ is the 4-velocity of Σ and − constructed as unitary theories. However, an important unan- ˙ ˙ swered question is how entanglement (and more general types nµ = ( R, U, 0,..., 0) (B2) − of quantum correlations) gets distributed between the tripartite system of gravity–early modes–late modes. Similarly, if the is its outward-pointingunit normal. The non-zero components fully formed horizon does not exist, it is important to investi- of the 4-acceleration are gate how (if at all) the soft hair properties of black holes [56] 0 ¨ 1 ′ ˙ 2 1 ¨ 1 ′ ˙ 2 ′ ˙ ˙ are modified. Finally, by studying a realistic models it is im- a = U F U , a = R+ (FU + F F ) U +F UR. − 2 2 portant to understand if absence of the event horizon leads to (B3) astrophysically significant differences with the classical col- Hence lapse. 3 1 K+ = R˙ U¨ U˙ R¨ F ′U˙ 2R˙ (F + F F ′) U˙ 3. (B4) ττ − − 2 − 2 U ACKNOWLEDGMENTS Calculating U¨ using Eq. (6) we obtain We thank Stefano Liberati and Bill Unruh for critical com- R¨ R˙ RF˙ ′ 1 ments, Paolo Pani, Tanmay Vachaspati and Yuki Yokokura for U¨ = 1 + U˙ useful suggestions. Robust discussions with Sabine Hossen- F F + R˙ 2 − ! F 2 F + R˙ 2 − ! felder are gratefully acknowledged. DRT and VB wish to FpU˙ 1 p thank Perimeter Institute and the Institute for Quantum Com- + U U˙ + . (B5) puting at the University of Waterloo for hospitality. VB is sup- F − 2 F + R˙ 2 ! ported by the Macquarie Research Fellowship scheme, RBM p was supported by the Natural Sciences and Engineering Re- The extrinsic curvature can be decomposed into the part into search Council of Canada. a part that does not involve the derivative FU (the classical part), and the part that is proportional to it (the evaporating + part), Kττ = c + e. We find Appendix A: Angular coordinates in D+1 dimensions K K 2R¨ + F ′ c = , (B6) We choose the convention for labelling angular coordinates K −2 R˙ 2 + F such that the generalization of the z-axis is three spatial di- p mensions is referred as x1, and there are D 1 angles φk. We while 0 − also introduce φ 0 and set ( ) 1. Then D ≡ 1 ∗ ≡ ˙ ˙ FU U 1 ˙ 1 ˙ 3 k−1 e = R U FU U Q K F 2 F + R˙ 2 − ! − 2 xk = r cos φk sin φl. (A1) 2 1 FU U˙ p Yl=1 = . (B7) −2 ˙ 2 The spherical part of the metric is F + R p D−1 a−1 The angular components — φkφk — of the extrinsic curva- 2 2 2 2 2 ture are calculated using the second relation in (B1). Starting r dΩD−1 := r dφ1 + sin φb dφa . (A2) a=2 ! with φ1φ1 — θθ in 3+1-D — we find X bY=1 a The coordinates y on the shell are (τ, φk), and the induced K+ = R˙ R + F U˙ R = R F + R˙ 2, (B8) metric is φ1φ1 p 2 2 2 ds = dτ + R (τ)dΩ − . (A3) where we have used the definition for U˙ , and Σ − D 1 F + R˙ 2 φ1 (B9) Appendix B: Thin shell collapse: outgoing Vaidya metric K+φ1 = , p R Dynamics of the shell is obtained via the second junction as expected. Going through a similar calculation we find that condition that equates the discontinuity of the extrinsic curva- Kφk = F + R˙ 2/R, for k =1,...,D 1. +φk − p 8

Using the appropriate deﬁnitions of extrinsic curvature for (u, r) that will transform the metrics into each other if C = the interior and the exterior regions, the components of the const. Starting from the metrics (3) and(4) we have 6 surface stress-energy tensor are ∂u 2 D−1 gtt = f(u, r) , (C1) τ 1 τ τ τ φk τ − ∂t S τ = K K− [K τ ]+ K φk h τ −8π +τ − τ − ∂u ∂u ∂u " k=1 ! # g = f(u, r) 1 0, (C2) X tr − ∂t ∂r − ∂t ≡ 1 F + R˙ 2 1+ R˙ 2 = (D 1) − = σ, ∂u 2 ∂u 8π − R − g = f(u, r) 2 . (C3) p p ! rr − ∂r − ∂r (B10) Hence we ﬁnd for the ττ component, and identical expressions such as ∂u 1 = , (C4) ¨ ′ ¨ ˙ 2 φ1 1 2R + F R 1 FU U ∂r −f(u, r) S φ1 = 8π " 2 F + R˙ 2 − 1+ R˙ 2 − 2 F + R˙ 2 and consequently grr = 1/f(u, r). To enforce gtt = p F + R˙ 2p 1+ R˙ 2 p f(u, r), +(D 2) − =0, (B11) − − R p p !# ∂u = 1, (C5) for all other components. ∂t ± should hold. However, the integrability condition implies

1. Classical shell dynamics ∂2u 1 ∂f ∂2u = = =0, (C6) ∂t∂r f 2(u, r) ∂t ∂r∂t The classical equation (9) is obtained by suppressing the time-dependence of the metric and integrating Eq. (B10). The that holds only in a stationary spacetime. shell surface density is conviniely expressed via the mass pa- rameter m, Appendix D: Thin-shell collapse: the retarded Schwarzschild m2 metric C =2m 1+ R˙ 2 . (B12) − RD−2 p The derivationproceeds analogouslyto case of the outgoing ˙ Fixing C for the initial condition R(0) = 0 we get Vaidya metric. The outward-pointing normal is

D−2 D−2 ˙ ˙ m = R 1 1 C/R , (B13) nµ = ( R, T, 0,..., 0), (D1) 0 − − 0 − q and the non-zero components of the 4-acceleration compo- where R(0) = R0. As a result, the equation of motion for R is nents are 1 2 t ¨ −1 −1 ˙ 2 ˙ ˙ ′ ˙ 2 dR C m a = T (τ)+ F F R +2R T F + FtT , = + 1. (B14) 2 t dτ − 2m 2RD−2 − h i(D2) s ′ r 1 −1 2 −1 ′ 2 1/(D−2) a = R¨ + F F R˙ +2R˙ T˙ F + F T˙ . The initial position can be speciﬁed as R0 = λC) , 2 t for some x> 1.Then h i(D3) m = λC(1 1 1/λ), (B15) Hence − − p 3 1 has the form that is independent of the dimension. In the limit K+ = R˙ T¨ T˙ R¨ + F −1F ′T˙ R˙ 2 F F ′T˙ 3 λ , m C/2. ττ − 2 − 2 → ∞ → 1 1 + F −1F R˙ T˙ 2 + F −1(F −1) R˙ 3 F (F −1) R˙ T˙ 2. 2 t 2 t − t Appendix C: Inequivalence between the retarded Schwarzschild (D4) and the outgoing Vaidya metrics Calculating T¨ using Eq. (5) we obtain Despite similar reasoning behind the two metrics (and the ′ ′ 3 ˜ 2R˙ RF¨ F F R˙ 2F R˙ Ft intuitive feeling that u and t are directly related), there is no ¨ ˙ 2 (D5) T = − − 3 (F +2R ). coordinate transformation between the coordinates (t, r) and 2F 2 F + R˙ 2 − 2F p 9

Substituting T¨ and with further simpliﬁcations we obtain Appendix E: Page’s formula in D spatial dimensions

2R¨ + F ′ F R˙ + t (D6) The ﬂux of radiation in D spatial dimensions is Kττ = + 2 . −2 F + R˙ 2 F D+1 J = σDT , (E1) The angular components ofp the extrinsic curvature are

where σD is the D-dimensional Stefan-Boltzmann constant. k−1 + + 1 rr ˙ 2 For comparison: Kφkφk = nφk;φk = 2 g gφkφk ,rnr = F (R,T )RT sin φb. b=1 2 Y (D7) π gDSD−2D!ζ(D) 1 σ3 = , σD = . (E2) µ µ 60~3c2 2(2π)D ~DcD−1 (note that ea = δa+1 for µ = 2,...D). From the above + equations one can therefore see that Kφkφk is not affected by the evaporation. The total radiated power is ˜ The derivatives of t are found via D+1 L = γσDTH AD, (E3) ∂ ∂t˜ ∂ r dr′ (t t˜)=1 = D−1 where γ is the numberof species, and AD = SD−1r , with ∂t − − ∂t ∂t R(t˜) f(t,˜ r) g Z the Schwarzschild radius 1 dR ∂t˜ r dr′ ∂f ∂t˜ = . 2 1 −F t,˜ R(t˜) dt˜ ∂t − R(t˜) f (t,˜ r) ∂t˜ ∂t D−2 Z rg = CD . (E4) (D8) Note that in 3D the constant σ3 is already calibrated to include In the limit r R(t˜) (and, i.e., T = t t˜) we have → → two polarizations of photons. The mass is given by ∂t˜ 1 dR ∂t˜ 1 = , (D9) − ∂t −F t,˜ R(t˜) dt˜ ∂t C (D 1)S − M = D − D 1 =: α C . (E5) 16πG D D hence (D+1)

∂t˜ 1 The horizon area is = . (D10) Σ D−1 ∂t 1 RT /F T, R(T ) D−1 D−2 − D−2 16πGD+1M AD = SD−1CD = SD−1 . (E6) Similarly, in the same limit we obtain (D 1)SD−1 − ∂t˜ 1 1 dR ∂t˜ The Hawking temperature dimensions is deﬁned as = . (D11) − ∂r Σ F T, R(T ) − F t,˜ R(t˜) dt˜ ∂r r=R(t) ′ f (rg) T = , (E7) Hence 4π

∂t˜ 1 which in D +1 spacetime dimensions is in = . (D12) ∂r Σ −F T, R(T ) Rt − 1 D 2 TD = D−2− . (E8) Changing to the coordinate time parametrization of the 4π √CD equation of motion requires the expressions for T˙ and T¨. From Eq. (3) we see that The rate of change of CD according to an observer at the spatial inﬁnity is dτ 2 = F dT 2 + R2 dT 2/F, (D13) T D − − D−2 dCD 16πG(D+1) 1 so = P (D) 2 (E9) dt − (D 1)S − D−2 − D 1 C −D/(D−2) −2/(D−2) −2/(D−2) F = α P (D)C =: ̟(D)C . ˙ (D14) D D − D T = 2 2 , sF RT − We will write the evaporation time and D D− 1 D 2 t = κC 2 κ = − , (E10) 2 2 E 0 dT˙ 2F RT RT T F + RT dF/dT ̟(D) D T¨ = T˙ = − T.˙ (D15) dT 2 F F 2 R2 where C is the initial value of C . − T 0 D q 10

a Appendix F: A thin shell collapsing on a massive core stress-energy tensor S b. The time-time component condition now reads

F + R˙ 2 F− + R˙ 2 τ 1 In the case of a thin shell collapsing on a core, the metric S τ = (D 1) − = σ, 8π − p Rq  − of the inner region is Schwarzschild as well, with f− = 1 D−2 − Cc/r . The equations of motion combine features of the   (F1) proceeding cases, and are calculated in terms of the surface while the angular component equations become

′ ′ ˙ 2 ˙ 2 1 2R¨ + F 2R¨ + F− F R˙ F + R F− + R t + (D 2) − =0. (F2) 8π  ˙ 2 − − F 2 − p Rq  2 F + R 2 F− + R˙ 2  p q 

Evaporation affects only the metric outside, i.e. and once R˙ & 2 | | R˙ 4 ′ C(D 2) dC ∂t˜ 1 ′ Cc(D 2) ¨ (F8) R 2 . F = − − , F− = −− , ≈−D(D 2κCx ) RD 1 − dt˜ ∂t RD 2 RD 1 − (F3) Similar to the case of a collapsing thin shell — without the where according to Eq. (E5) core — we have that

C = M/α , C = M /α . (F4) dR dR R˙ D c c D R˙ = T˙ . (F9) dt ≈− dt F Since in this regime On the other hand, only the outside term ∂t˜ 1 ∂t˜ 1 R˜ Σ = Rt RT F, , = . dC ∂t˜ 1 t| ≡ ≈− ∂t ≈ 2 ∂r −2F F = , (F5) Σ Σ t D−2 (F10) − dt˜ ∂t R The evaporation equation in D +1 -dim (as seen from Alice) enters the equation. The qualitative behaviour of the collaps- dC dC ∂t˜ ∂t˜ ing shell is the same as that for a collapsing shell with no core. = T˙ + R˙ , (F11) dτ dt˜ ∂t ∂r The only effect of the core on the dynamics of the system is to accelerate the rate of evaporation. can be approximated as

dC R˙ 1 R˙ C˙ = . (F12) ≈− dt˜ F Dκ C1/(D−2) 1. Asymptotic dynamics analysis Eqs (F8) and (F12) give us the dynamics of the shell and of the Schwarzschild radius. Hence We again consider x = R rg. By expanding R¨ in inverse powers of C and x, and only− taking the leading terms that 1 C˙ 1 R˙ r˙g = = . (F13) diverge when x 0 D 2 C(D−3)/(D−2) D(D 2)κ Cx → − − that means the gap x evolves according to ˙ 2 ˙ 2 ¨ R R +1 −1 R + x , (F6) 1 1 1 1 ≈ 2D(D 2)κx2pC R˙ + 1+ R˙ 2 O x˙ = R˙ 1 = R˙ 1 , − − D(D 2)κ Cx | | D(D 2)κ Cx − p − − (F14) the dominant terms of the acceleration are the same as for when x<ǫ∗, with Eq. (36). We can still use the approximations 1 ǫ∗(τ)= . (F15) D(D 2)κC(τ) (D 2)x (D 2)x ˙ rg ˙ − F 1/−(D−2 − , T R, ≈ C ) ≡ rg ≈−(D 2)x we are guaranteed x˙ > 0, i.e. the shell does not cross the − (F7) Schwarzschild radius. 11

[1] L. D. Landau and E. M. Lifshitz, Classical Theory of Fields [28] H. M. Haggard and C. Rovelli, Phys. Rev. D. 92, 104020 (Reed International, Oxford, 1975). (2015). [2] C. Kiefer, Quantum Gravity, (Oxford University Press, Oxford, [29] V. Baccetti, V. Hussain, and D. R. Terno, Entropy 19, 17 (2017). 2007). [30] V. Baccetti, R. B. Mann, and D. R. Terno, arXiv:1703.09369 [3] N. D. Birrel and P. C. W. Davies, Quantum Fields in Curved (2017). Space (Cambridge University Press, Cambridge, 1986). [31] E. Poisson, A Relativist’s Toolkit, (Cambridge University Press, [4] R. B. Mann, Black Holes: Thermodynamics, Information, and Cambridge, 2004). Firewalls (Springer, New York, 2015). [32] G. T. Horowitz, ed., Black Holes in Higher Dimensions, (Cam- [5] D. Harlow, Rev. Mod Phys. 88, 015002 (2016). bridge University Press, Cambridge, 2012). [6] S. W. Hawking, Nature 248, 30 (1974). [33] W. Israel, Nuovo Cimento 44B, 1 (1966); 48B, 463 (1967). [7] A. D. Heffler, Rep. Prog. Phys. 66, 943 (2003). [34] P. C. W. Davies, S. A. Fulling, and W. G. Unruh, Phys. Rev. D [8] D. N. Page, Phys. Rev. D 13, 198 (1976). 13, 2720 (1976). [9] A. Ashtekar and M. Bojowald, Class. Quant. Grav 22, 3349 [35] P. Háj´ıˇcek, Nucl. Phys. B 603, 555 (2001); M. Ambrus and P. (2005). Háj´ıˇcek, Phys. Rev. D 72, 064025 (2005). [10] R. M. Wald, Living. Rev. Rel. 4, 6 (2001). [36] A. Saini and D. Stojkovic, Phys. Rev. Lett. 114, 111301 (2015). [11] S. D. Mathur, Class. Quantum. Grav. 26, 224001 (2009). [37] P. C. Vaidya, Phys. Rev. 83, 10 (1951). [12] W. G. Unruh and R. M. Wald, arXiv:1703.02140 (2017); D. [38] M. K. Parikh and F. Wilczek, Phys. Lett. B 449, 24 (1999). Marolf, arXiv:1703.02143 (2017). [39] F. Fayos and R. Torres, Class. Quant. Grav. 25, 175009 (2008). [13] M. Visser, PoS BHs,GRandStrings 2008:001 (2008), [40] D. Page and W. H. Zurek, Phys. Rev. D 29, 628 (1984). arXiv:0901.4365v3. [41] H. D. Zeh, The Nature and Origins of Time-Asymmetric Space- [14] U. H. Gerlach, Phys. Rev. D 14, 1479 (1976). time Structures, in A. Ashtekar and V. Petkov (eds.), Springer [15] P. Háj´ıˇcek, Phys. Lett. B 182, 309 (1986); P. Háj´ıˇcek, Phys. Rev. Handbook of Spacetime (Springer, Berlin, 2014), p. 185. D 36 1065 (1987). [42] L. C. Barbado, C. Barceló, and L. J. Garay, Class. Quant. Grav. [16] G. L. Alberghi, R. Casadio, G. P. Vacca, and G. Venturi, Phys. 28, 125021 (2011). Rev. D 64, 104012 (2001). [43] H. Kawai and Y. Yokokura, Phys. Rev. D. 93, 044011 (2016). [17] C. Barceló, S. Liberati, S. Sonego, M. Visser, Class. Quant. [44] R. Brustein, Fortschr. Phys. 62, 255 (2014). Grav. 23, 5341 (2006); C. Barceló, S. Liberati, S. Sonego, M. [45] R. Narayan and J. E. McClintock, New Astr. Rev. 51, 733 Visser, Phys. Rev. D 77, 044032 (2008). (2008); M. Abramowicz and C. Fragile, Living Rev. Relativ- [18] T. Vachaspati, D. Stojkovic, and L. M. Kraus, Phys. Rev. D 76, ity 16, 1 (2013). 024005 (2007); T. Vachaspati and D. Stojkovic, Phys. Lett. B [46] V. Cardoso, E. Franzin, and P. Pani, Phys. Rev. Let. 116, 171101 663, 107 (2008). (2016); V. Cardoso, S. Hopper, C. F. B. Macedo, C. Palenzuela, [19] H. Kawai, Y. Matsuo, and Y. Yokokura, Int. J. Mod. Phys. A 28, and P. Pani, Phys. Rev. D 94, 084031 (2016). 1350050 (2013); H. Kawai and Y. Yokokura, Int. J. Mod. Phys. [47] C. R. Stephens, G. ’t Hooft, and B. F. Whiting, Class. Quant. A 30, 1550091 (2015). Grav. 11, 621 (1994). [20] P.-M. Ho, Nucl. Phys. B 909, 394 (2016). [48] S. L. Braunstein, S. Pirandola, and K. Zyczkowski,˙ Phys. Rev. [21] S. W. Hawking, Phys. Rev. D 14, 2460 (1976). Lett. 110, 101301 (2014). [22] D. R. Terno, Quantum Information and Relativity Theory, Ph.D. [49] S. Lloyd, and J. Preskill, JHEP 1408, 126 (2014). thesis, (Technion, Haifa, 2003). [50] J. Maldacena, and L. Susskind, Fortschr. Phys. 61, 781 (2013). [23] A. Peres and D. R. Terno, Rev. Mod. Phys. 76, 93 (2004). [51] J. C. Baez and J. Vicary, Class. Quant. Grav. 31, 214007 (2014). [24] A. Dragan, arXiv:1610.07839 (2010). [52] S. Abdolrahimi, D. N. Page, and C. Tzounis, arXiv:1607.05280 [25] J. F. Donoghue, Phys. Rev. D 50, 3874 (1994); C.P. Burgess, (2016). Living Rev. Relativity 7, 5 (2004). [53] P.-M. Ho, Class. Quant. Grav. 34, 085006 (2017). [26] B.-L. Hu and E. Verdauger, Class. Quant. Grqav 20, R1 (2003); [54] B. S. Kay, Class. Quant. Grav. 15, L89 (1998); B.-L. Hu and E. Verdauger, Liv. Rev. Relativity 11, 3 (2008). arXiv:hep-th/9802172 (1998). [27] A. Almhieri, D. Marolf, J. Polchinski, and J. Sully, JHEP 02, [55] V. Husain and D. R. Terno, Phys. Rev. D 81, 044039 (2010). 062 (2013). [56] S. W. Hawking, M. J. Perry, and A. Strominger, Phys. Rev. Lett. 116, 231301 (2016).